Estimation of the prevalence of adverse drug reactions from social media

Highlights • Word2vec was utilized to discover variants of adverse drug reaction (ADR) terms in social media data. • The prevalence of ADR derived from social media using original lexicon has a decent correlation with that from official sources. • The lexicon derived by word2vec improved the correla...

Full description

Saved in:
Bibliographic Details
Published in:International journal of medical informatics (Shannon, Ireland) Ireland), 2017-06, Vol.102, p.130-137
Main Authors: Nguyen, Thin, Larsen, Mark E, O’Dea, Bridianne, Phung, Dinh, Venkatesh, Svetha, Christensen, Helen
Format: Article
Language:English
Subjects:
Adverse drug reactions
Australia - epidemiology
Consumer health informatics
Databases, Factual
Drug informatics
Drug-Related Side Effects and Adverse Reactions - epidemiology
Drug-Related Side Effects and Adverse Reactions - prevention & control
Humans
Internal Medicine
Machine Learning
Other
Prevalence
Social media
Social Media - utilization
Word embedding
Word representation
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Highlights • Word2vec was utilized to discover variants of adverse drug reaction (ADR) terms in social media data. • The prevalence of ADR derived from social media using original lexicon has a decent correlation with that from official sources. • The lexicon derived by word2vec improved the correlation. • Advanced cluster computing was employed to process 6.4 terabytes of data containing 3.8 billion records.
ISSN:1386-5056
1872-8243
DOI:10.1016/j.ijmedinf.2017.03.013